The chaperone heat shock protein 90 (Hsp90) is an emerging target in cancer treatment due to its important roles in maintaining transformation and in increasing the survival and growth potential of cancer cells. Hsp90 function is regulated by a pocket in the N-terminal region of the protein that binds and hydrolyzes ATP. Occupancy of this pocket by high affinity ligands prevents the dissociation of Hsp90 client proteins from the chaperone complex and as a consequence, the trapped proteins do not achieve their mature functional conformation and are degraded by the proteasome. Protein clients of Hsp90 are mostly kinases, steroid receptors and transcriptional factors involved in driving multistep-malignancy and in addition, mutated oncogenic proteins required for the transformed phenotype. Examples include Her2, Raf-1, Akt, Cdk4, cMet, mutant p53, ER, AR, mutant BRaf, Bcr-Abl, Flt-3, PoIo-I kinase, HQF-I alpha and hTERT. Degradation of these proteins by Hsp90 inhibitors leads to cell-specific growth arrest and apoptosis in cancer cells in culture, and to tumor growth inhibition or regression in animal models. One such inhibitor, 17-allyl-amino-desmethoxy-geldanamycin (17 AAG, FIG. 1A) has entered clinical trials in cancer patients in the US and UK and has shown early evidence of therapeutic activity when administered alone or in combination with docetaxel. Despite these early promising results, 17AAG has several potential limitations. Most prominent are its limited solubility and cumbersome formulation. It also exhibits dose and schedule dependent liver toxicity believed to be caused by the benzoquinone functionality. Radicicol (RD, FIG. 1B) a structurally unrelated natural product, has biological activity similar to that of 17AAG but is not hepatotoxic, yet no derivative of this class has made it into clinic.
Making use of the peculiar bent shape of Hsp90 inhibitors and of existent Hsp90 crystal data, purine-scaffold derivatives with Hsp90 inhibitory activities have been designed. PCT Application No. WO02/36075, which is incorporated herein by reference identified a generalized structure for purine scaffold inhibitor class of compounds with the formula:

The first synthesized derivative of this class, PU3 (FIG. 1C), bound Hsp90 with moderate affinity and elicited cellular effects that mimic 17AAG addition. Preliminary efforts focused at improving the potency of this agent have mostly focused on modifying the left side of the scaffold and have led to the synthesis of several compounds with improved activity in both biochemical and cellular assays. One such compound, PU24FC1 (FIG. 1D) is a potent and selective inhibitor of tumor Hsp90 and exhibits anti-tumor activities in both in vitro and in vivo models of cancer. Other purine-scaffold compounds with higher potency over PU24FC1 in in vitro models of cancer have subsequently been disclosed. (See, PCT Patent Publication No. WO03/037860, which is incorporated herein by reference.) Although a significant number of derivatives has been created by these combined efforts, the nature and position of substituents on the right side aryl moiety (X2 in the structure above) and nature of the linker between the two rings have not been sufficiently investigated. The present invention provides a class of Hsp90 inhibitors with enhanced activity as compared to previously known compounds and a class of inhibitors with differential selectivity and activity for Rb normal versus Rb defective cells.